def getCallbacks(patience_es, patience_rlr):
"""
FUNCTION:
Returns a list of callbacks with the provided properties
PARAMS:
patience_es: int
Number of iterations to wait before EarlyStopping is performed
patience_rlr: int
Number of iterations to wait before ReduceLearningRate is performed
RETURNS:
List of callbacks
"""
return modelFuncs.getBasicCallbacks(patience_es=patience_es, patience_rlr=patience_rlr)
def setLSTMCallbacks(patience_es, patience_rlr):
"""
FUNCTION:
Redefines the default LSTM callbacks
NB: only for current state
PARAMS:
patience_es: int
Number of iterations to wait before EarlyStopping is performed
patience_rlr: int
Number of iterations to wait before ReduceLearningRate is performed
RETURNS:
None
"""
global _default_LSTM_args
_default_LSTM_args['callbacks'] = modelFuncs.getBasicCallbacks(patience_es=patience_es, patience_rlr=patrience_rlr)
_X_train = None
_y_train = None
_X_test = None
_y_test = None
_maxEnrolWindow = None
_indexColumn = None
_default_MLP_args = {
'activation': 'relu',
'loss': 'mean_absolute_error',
'optimizer': 'adam',
'metrics': ['mean_absolute_error'],
'epochs': 500,
'batchSize': 128 * 2,
'verbose': 1,
'callbacks': modelFuncs.getBasicCallbacks(patience_es=60, patience_rlr=40),
'enrolWindow': 0,
'validationSize': 0.2,
'testSize': 0.2,
}
_default_LSTM_args = {
'activation': 'tanh',
'loss': 'mean_absolute_error',
'optimizer': 'adam',
'metrics': ['mean_absolute_error'],
'epochs': 500,
'batchSize': 128 * 2,
'verbose': 1,
'callbacks': modelFuncs.getBasicCallbacks(patience_es=60, patience_rlr=40),
'enrolWindow': 12,
from howiml.utils import plots
from howiml.utils import prints
from howiml.utils import analysis
import numpy as np
import tensorflow as tf
_default_MLP_args = {
'activation': 'relu',
'loss': 'mean_squared_error',
'optimizer': 'adam',
'metrics': ['mean_squared_error'],
'epochs': 2000,
'batchSize': 64,
'verbose': 1,
'callbacks': modelFuncs.getBasicCallbacks(patience_es=300, patience_rlr=150),
'enrolWindow': 0,
'validationSize': 0.2,
'testSize': 0.2,
}
_default_LSTM_args = {
'activation': 'tanh',
'loss': 'mean_squared_error',
'optimizer': 'adam',
'metrics': ['mean_squared_error'],
'epochs': 500,
'batchSize': 128,
'verbose': 1,
'callbacks': modelFuncs.getBasicCallbacks(patience_es=75, patience_rlr=50),
'enrolWindow': 32,